All pull requests

New pull request

[DNM] test SMG aggregated vLLM through srt-slurm
#3660cquil11 opened Oct 2, 2026Collaborator
1
GLM-5.2 NVFP4 GB300 SGLang AgentX: nightly-dev-20261002, MTP 5-1-6 aggregate, drop no-op settings / GB300 GLM-5.2 NVFP4 SGLang AgentX：更新 nightly、聚合 MTP 5-1-6、删除无效设置
#3659nvpohanh opened Oct 2, 2026Collaborator5 of 11 tasks
1
Add GB300 partition routing under one cluster profile
#3658cquil11 opened Oct 2, 2026Collaborator
1
Add GLM-5.3 FP8 SGLang AgentX on MI300X with MoRI and UMBP
#3657cquil11 opened Oct 2, 2026Collaborator
1
Add GLM-5.3 MI325X AgentX with MORI and UMBP
#3656cquil11 opened Oct 2, 2026Collaborator
1
[DNM] test SMG aggregated SGLang through srt-slurm
#3655cquil11 opened Oct 2, 2026Collaborator
1
Update B300 DSV41flash vLLM image and retune TP4/DEP2 / 更新 B300 DSV41flash vLLM 镜像并调优 TP4/DEP2
#3653Juntian777 opened Oct 2, 2026Collaborator7 of 11 tasks
3
Update GB300 DSV41flash vLLM image and retune TP4/DEP2 / 更新 GB300 DSV41flash vLLM 镜像并调优 TP4/DEP2
#3652Juntian777 opened Oct 2, 2026Collaborator7 of 11 tasks
2
fix(klaud): defer unrunnable baseline rosters / 延后无法运行的基线候选
#3643adibarra opened Oct 1, 2026Collaborator
Single-node vLLM op-attribution profiling: per-kernel op/module, clocks, MoE routing
#3639hbarclay opened Oct 1, 2026Collaborator
[AMD][MI355X] Kimi-K3 MXFP4 vLLM: bump image to nightly-rocm100-ac9126e / Kimi-K3 MXFP4 vLLM：镜像升级至 nightly-rocm100-ac9126e
#3635seungrokj opened Oct 1, 2026Collaborator1 task
4
[PowerX] Kimi-K3 B300 AgentX measured-power coverage
#3632edwingao28 opened Oct 1, 2026Collaborator7 of 10 tasks
1
Add B300 Dynamo+SGLang AgentX configs for DeepSeek V4 with aggregate DEP8 c384 / 添加含聚合式 DEP8 c384 的 DeepSeek V4 B300 Dynamo+SGLang AgentX 配置
#3631nvpohanh opened Oct 1, 2026Collaborator
6
feat(agentx): refresh DSV4 GB300 SGLang on nightly-dev-20261001 / 基于 nightly-dev-20261001 更新 DSV4 GB300 SGLang 配置
#3630nvpohanh opened Oct 1, 2026Collaborator
4
Add DSV4 GB200 Dynamo+SGLang AgentX configs without W4A4 MegaMoE / 添加不启用 W4A4 MegaMoE 的 DSV4 GB200 Dynamo+SGLang AgentX 配置
#3629nvpohanh opened Oct 1, 2026Collaborator
3
[PowerX] MiniMax-M3 B200 AgentX measured-power coverage
#3628edwingao28 opened Oct 1, 2026Collaborator6 of 10 tasks
5
[AMD][MI355X] Kimi-K3 vLLM AgentX FP8 per-channel linear quantization-config / [AMD][MI355X] Kimi-K3 vLLM AgentX 新增 FP8 per-channel linear quantization-config
#3627seungrokj opened Oct 1, 2026Collaborator
2
feat(srt): check srt-slurm recipes against master images before sweep dispatch / feat(srt)：在派发 sweep 前检查 srt-slurm 配方与主配置镜像是否一致
#3624chunfangamd opened Oct 1, 2026Collaborator5 of 10 tasks
1
[Klaud Cold] [TileRT] GLM-5.3 FP8 MI355X AgentX: tilert 0.1.6.post3 + vLLM 0.28/ATOM prefill / [Klaud Cold] [TileRT] GLM-5.3 FP8 MI355X AgentX：tilert 0.1.6.post3 + vLLM 0.28/ATOM prefill
#3623functionstackx opened Oct 1, 2026Collaborator5 of 10 tasks
3
[GLM-5.2] Enable iter perf stats and metric publishing on GB300 disagg / [GLM-5.2] 为 GB300 disagg 启用迭代性能统计和指标发布
#3621richardhuo-nv opened Sep 30, 2026Collaborator
4
Enable MiniMax-M3 H100 FP8 indexer cache
#3620RohitNagraj opened Sep 30, 2026Collaborator
2
[TileRT] GLM-5.3 FP8 MI355X AgentX: tilert 0.1.6.post3 + vLLM 0.28/ATOM prefill / [TileRT] GLM-5.3 FP8 MI355X AgentX：tilert 0.1.6.post3 + vLLM 0.28/ATOM prefill
#3617CrimsonDump opened Sep 30, 20265 of 10 tasks
[AMD] Update GLM-5.2 MI355X image to 20260930 daily / [AMD] 更新 GLM-5.2 MI355X 镜像至 20260930 daily
#3616jiejingzhangamd opened Sep 30, 2026Collaborator22 of 32 tasks
3
fix(launch): give single-node srun steps their GPUs' share of node DRAM
#3615cquil11 opened Sep 30, 2026Collaborator
3
feat(launch): re-add llm-d multinode support to infx.launch
#3613adibarra opened Sep 30, 2026Collaborator
1